You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by "Karl Wettin (JIRA)" <ji...@apache.org> on 2008/04/11 16:16:06 UTC
[jira] Resolved: (MAHOUT-26) Matrix implementation bug fix and
little addition
[ https://issues.apache.org/jira/browse/MAHOUT-26?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Karl Wettin resolved MAHOUT-26.
-------------------------------
Resolution: Fixed
Laplace parts commited in reversion 647148.
Thanks Sergey!
> Matrix implementation bug fix and little addition
> -------------------------------------------------
>
> Key: MAHOUT-26
> URL: https://issues.apache.org/jira/browse/MAHOUT-26
> Project: Mahout
> Issue Type: Bug
> Environment: JDK 1.6, IntelliJ Idea 7.0.2
> Reporter: Sergey Chickin
> Assignee: Karl Wettin
> Attachments: MAHOUT-26.patch, MAHOUT-26a.patch
>
>
> Some bugfixes to matrix multiplication and implementation of recursive determinant calculation using Laplace theorem
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
Re: [jira] Resolved: (MAHOUT-26) Matrix implementation bug fix and
little addition
Posted by Karl Wettin <ka...@gmail.com>.
Ted Dunning skrev:
>
> Laplace's method is, btw, one of the worst ways to compute determinants. It
> is on a par with using Cramer's rule for solving linear systems.
>
> Much better to do QR and take the product of the diagonal of the R matrix or
> do the same with the diagonal matrix out of an SVD. Even LU decomposition
> is better.
>
> It is nice to have implementations like this and have people exercise the
> matrix code, but nobody should actually be using these algorithms for
> anything serious.
At first I was not sure to add it as there are no use cases yet, but
then I thought it's nice with some stuff in the trunk.
karl
Re: [jira] Resolved: (MAHOUT-26) Matrix implementation bug fix
and little addition
Posted by Ted Dunning <td...@veoh.com>.
Laplace's method is, btw, one of the worst ways to compute determinants. It
is on a par with using Cramer's rule for solving linear systems.
Much better to do QR and take the product of the diagonal of the R matrix or
do the same with the diagonal matrix out of an SVD. Even LU decomposition
is better.
It is nice to have implementations like this and have people exercise the
matrix code, but nobody should actually be using these algorithms for
anything serious.
On 4/11/08 7:16 AM, "Karl Wettin (JIRA)" <ji...@apache.org> wrote:
>
> [
> https://issues.apache.org/jira/browse/MAHOUT-26?page=com.atlassian.jira.plugin
> .system.issuetabpanels:all-tabpanel ]
>
> Karl Wettin resolved MAHOUT-26.
> -------------------------------
>
> Resolution: Fixed
>
> Laplace parts commited in reversion 647148.
>
> Thanks Sergey!
>
>> Matrix implementation bug fix and little addition
>> -------------------------------------------------
>>
>> Key: MAHOUT-26
>> URL: https://issues.apache.org/jira/browse/MAHOUT-26
>> Project: Mahout
>> Issue Type: Bug
>> Environment: JDK 1.6, IntelliJ Idea 7.0.2
>> Reporter: Sergey Chickin
>> Assignee: Karl Wettin
>> Attachments: MAHOUT-26.patch, MAHOUT-26a.patch
>>
>>
>> Some bugfixes to matrix multiplication and implementation of recursive
>> determinant calculation using Laplace theorem